Diurnal and nocturnal sleep stage patterns following sleep deprivation· RALPH J. BERGER, JAMES M. WALKER, THOMAS D. SCOTT LINDA J. MAGNUSON and STEPHEN L. POLLACK Division of Natural Sciences I, University of California, Santa Cruz, Calif. 95060 The extent to which the propensities for Stage 4 and REM sleep are dependent upon amount oe sleep loss, length of prior wakefulness, and circadian faetors was examined. Diurnal and nocturnal sJeep stage patterns in eight male and four female Ss were studied following 24 and 36 h of sleep deprivation, respectively. Stage 4 inereased on day (p< .01) and night (p < .001) recovery but remained eonfined to the first third of sleep on both conditions. REM lateney was reduced on day recovery (p< .05), but percentage REM did not change following sleep deprivation. The propensity for Stage 4 appears to be influenced by the amount of sleep loss and that for REM by circadian factors.

these served as adaptation nights, and the third was used as a baseline of their normal nocturnal sleep patterns. On two other occasions, Ss stayed awake for an entire night while monitoring sleep recordings in the laboratory as part of an upper division psychology course requirement. On one of these occasions they went to bed at 9:00 a.m. and on the other, the following night, at 12:00 a.m., thereby undergoing 24 and 36 h of sleep deprivation, respectively. The order of the two sleep-deprivation conditions was counterbalanced among Ss. Eleetrodes were placed for recording eJectroencephalograms, chin electromyograms, and electrooculograms, and the records were coded and scored blind for s1eep stages, by 30-sec epochs, aecording to standard procedures (Rechtschaffen & KaIes, 1968). On all occasions, recordings were taken for aperiod of 8 h. following whieh Ss were awakened. Comparisons between sleep eonditions were made by one-way analysis of variance for repeated measurements (Winer, 1962). Duncan's multiple range test was used to compare speeifie means. Hourly distribution of sleep stages and wakefulness were compared with t tests for corretated means. RESULTS Patterns of sleep on recovery following both s1eep-deprivation conditions were similar in many respects but differed from baseline sleep. In other respeets, the recovery conditions were different from each other, as weil as from baseline s1eep.

Rapid eye movement (REM) and nights following partial or total sJeep Stage 4 sleep are differentially deprivation (Berger & Oswald, 1962; distributed within uninterrupted Kales, Tan, Konar, Naitoh, Preston, & nocturnal sleep (Dement & Kleitman, Malmstrom, 1970; Webb & Agnew, 1957a, b; Williams, Agnew, & Webb, 1965; Williams, Hammack, Daly, 1964, 1966). Stage 4 predominates Dement, & Lubin, 1964). This study du ring the first third of the night, was designed to differentiate between whereas REM increases progressively, the relative influences of these three reaching a plateau during the last third faetors by examining diurnal and of the night, so that it remains nocturnal sleep-stage patterns on constant even when sleep is extended recovery from 24 and 36 h of s1eep for longer than usual (Phoebus, Taub, deprivation, respectively. METHOD Globus, & Drury, 1970; Verdone, The Ss were eight male and four 1968). If naps are taken subsequent to a female college students who had heen regular night of sleep, the relative instructed to maintain their regular distribution of REM and Stage 4 sleep habits and to have abstained c hanges as the naps are taken from drugs, including alcohol. The Ss were studied in pairs, progressively Jater in the day (Maron, Rechtschaffen, & Wolpert, 1964; sleeping in separate bedrooms, on five Webb & Agnew, 1967; Webb, Agnew, different occasions, separated by at & Sternthai, 1966). Naps taken early least 1 week. On three nights, Ss in the morning resemble a retired at 12:00 a.m.; the first two of continuation of the terminal portion of the previous night's sJeep, with large Table 1 amounts of REM and negligible Minute. of Total Sleep Time, Initial Wakefulne.., REM Latenc:y,and Lenlth 01 REM amounts of Stage 4. Naps taken in the Cyc!e, and Number of REM Pedoda evening resemble the first portion of Nl&ht Recovery Oay Recovery Baseline night sleep, with large amounts oe Stage 4 and small amounts of REM. Mean SO Mean SO Mean SO Naps taken midday or in the early 431.7* 46.4 427.8** 32.0 8.1 468.5 afternoon have an intermediate Total Sleep Time 11.9 7.2 13.0 6.5 17.4t 9.9 Initial Wakefulness pattern. Latency 110.4 50.0 105.1 45.1 70.ott 31.5 Are the propensities for REM and REM Cycle 87.2 101.5 19.8 14.0 Leugth 88.2 12.6 Stage 4 sleep influeneed by the length REM 1.00 4.92 4.08t 0.90 4.75 0.62 REM Perioda of time aperson has been awake sinee his major sleep period, by how much tp < .05-Signi(icantly different (rom day and night recovery. of his normal sleep he might have lost, ttp < .05-Signi(icantly different (rom ba.seline and night recovery. by preexisting circadian rhythms, or *p < .05, **p < .Ol-Signi(icantly different (rom night recovery. by a combination of these three Table 2 faetors? Earlier work indicates that Percentaae of Total Sleep Time Spent in Each Sleep Stage amounts of Stage 4 are dependent Baseline nay Recovery Ni&ht Recovery upon one or both of the first two factors, ;since Stage 4 is increased on *The assistance of Joseph AveUar, Cecil Burchfiel, Anita Harten, James Rossie, Willard Snow, Marcum Sproule, Christine Summers. Susan Thatcher, Thomas Tower, and Fredric Wiebe is gratefully acknowledged. Supported in part by National Science Foundation Grant GB-8782.

Psychon. Sei., 1971, Vol. 23 (4)

Sleep Stage

Mean

sn

Mean

sn

Mean

sn

1 2 3 4 REM

13.0 55.0 10.0 2.5 19.5

4.5 7.2 3.7 3.3 4.7

13.4 53.0 10.7 5.1* 17.8

4.0 6,2 3.4 6.8 5.0

7.8t 55.7 13.3 5.8** 17.4

5.0 5.9 5.5 6.1 4.8

*p tp

< <

. 01, **p < .001-Signi(icantly different (rom baseline . .Ol-Significantly different (rom baseline and day recovery.

273

recovery indicates a persisting innuence of the normal circadian distribution of REM sleep, in spite of 60sleep deprivation. However, the amounts of REM sleep in the first few hours of day recovery sleep were no higher than in the other two conditions, which suggests that Stage 4 sleep takes priority on recovery from s1eep deprivation independently of preexisting circadian innuences. Weitzman, Kripke, Goldmacher, McGregor, & Nogeire (1970) recently described sleep patterns following acute reversal of the sleep'waking cycle after 1 night without sleep. · A1though their. data on daytime sleep was computed over 3·day periods 40 rather than for each individual s1eep period, their results are generally L simitar to these. After reversal, diurnal 1\ sleep was characterized by shorter I~ REM latencies, shorter episodes of Stage 2, and more wakefulness in the lallt third of sleep than. prior to 4 1& 7 e HOUR OF $1. EEP P(RIOO reversal. Stage 4. predominated in the first ·third of the sleep per iod before Fig. 1. Percentage of each IlUcceaive hour apent awake and in ~ch sleep stage and after reversal . On the o,ther hand, we did not confirm their findihgs of an under each condition. increased number of stage changes, decreased duration of REM per iods, or (p < .05, Table 1), but percentage increased number of interruptions of Similarities Between REM did not change following either REM periods during daytime sleep. Recovery Conditions The Ss fell asleep more rapidly, with sleep-deprivation condition (Table 2). Using the same criteria for defining a less initial wakefulness (p < .05) prior REM period, as successive episodes of DISCUSSION to first entering Stage 2, on both REM separated by less than The increase in Stage 4 following 15 min, no differences were found in recovery conditions (Table 1). Percentageof Stage 4 was higher on both sleep-deprivation conditions the number of interrupted REM both day recovery (p < .01) and night confirms earlier findings (Berger & periods, total number of interruptions, recovery (p < .001) than on baseline Oswald, 1962; Kaies et, al, 1970; or in the proportion of total number nights (Table 2). This was especially Williams et aI, 1964). Ss lost one of REM periods which were true for the first hour of sleep night's sleep on both sleep-deprivation interrupted ' between the three (p < . 001, Fig. 1) and renected conditions, but amounts of Stage 4 did conditions. Possibly, the changes quicker entry into Stage 4 as weil as not differ between the two conditions, Weitzman et aI described were not greater amounts of Stage 4 . even though Ss had been awake for .present on the first reversal day of The total number of REM periods 36 h on the night recovery condition sleep in their study hut developed only was .greater on both recovery as compared with 24 h on the day on the second or third day, so that conditions than on baseline nights recovery condition. It, thereCore, their 3-day block of data is not strictly (p< .05) and renected decreases in appears that the propensity for Stage 4 comparable with the day recovery the length oC the REM cycle which is primarily determined by the amount condition of this study. just failed to reach statistical oC habitual sleep a person has lost rather than the total time he has BERGER, R. REFERENCES significance (p < .10, Table 1). J .• &. OSWALD, I. Effects of remained awake. The distribution of Differences Between sleep deprivation on behaviour, Stage 4 also remained independent of Recovery Conditions subsequent sleep. and dreaming. Journal of Mental Science, 1962, 108, 457-465. Total time spent asleep was greater the normal circadian rhythm, being DEMENT, W., &. KLEITMAN, N. Cyclic on night recovery than on baseline largely confined to the first third of variations in EEG during sleep and their nights (p < .05) and day recovery sleep under all conditions. relat ion to eye movements. bodY motility, and dreaming. By contrast, amounts of REM sleep (p< .01, Table 1). S1eep on night Electroencephaloeraphy &: Clinical recovery was sounder, with leu did not differ among the three Neurophysiolo&Y, 1967a, 9, 673-690. Stage 1 (p < .01, Table 2) and fewer conditions. In an earlier study, REM DEMENT, W., .. KLEITMAN, N. The transitions to Stage 1 from other sleep was reduced during noctumal sleep relation of eye movements durinll sleep to dream activity: An obiective method for stages (X= 24.8, p< .05), than on aCter 108 h oC total sleep deprivation the study of dreams. Journal of baseline nights (X= 35.0) and day (Berger & Oswald, 1962). The absence Experimental Psychology, 1957b, 53, recovery (X= 39.6). Ss tended to wake of a simitar reduction in the present 339-346. up earlier on day recovery than on study may be due to the shorter length KALES, A ., TAN. T. L., KOLLAR. E. J .• NAITOH, P., PRESTON, T. A., .. night recovery so that wakefulness was of s1eep deprivation, although extreme MALMSTROM, E. J. Sieep patterns increased in the seventh (p < .05) and lengths of deprivation of up to 205 h foUowing 205 hour! of sleep deprivation. eighth (p < ,05) hours of the recording have been reported to lead to increases Psychosomatic Medicine. 1970, 32, 189-200. in REM sleep (Kaies et al, 1970). session (Fig. 1). L., RECHTSCHAFFEN. A., .. Latency to the first REM period Nevertheleu, the distribution of REM MARON, WOLPERT. E . A. Sieep cycle during was shorter on day recovery than on sleep differed between the conditions. nappinll. Archives of General Psychiatry, 1964, 11,503-508. baseline nights or night recovery The shortened REM latency on day S T~GE

STAG [

2

~o

t~ fr<"~,.0 ~

274

Psychon. Sei., 1971, Vol. 23 (4)

PHOEBUS, E., TAUB, J. M., GLOBUS, G. G., & D R U R Y , R. Sleep stage characteristics of extended sleep in normal subjects. Perceptual & Motor Skills, 1970, 31,526. RECHTSCHAFFEN, A., & KALES, A. (Eds.) A manual of standardized terminology, techniques, and seoring system for sleep stages of human subjects. Washington, D.C: U.S. Government Printing Office, 1968. VERDONE, P. Sleep satiation: Extended sleep in normal subjects. Electroencephalography & CUnical Neurophysiology, 1968, 24, 417-423. WEBB, W. B., & AGNEW, H. W., JR. Sleep: Effects of a restricted regime. Science, 1965,150,1745-1747. WEBB, W. B., & AGNEW, H. W., JR. Sleep cycling within twenty-four hour periods. Journal of Experimental Psychology, 1967,74,158-160. WEBB, W. B., AGNEW, H. W., JR., & STERNTHAL, H. Sleep during the early morning. Psychonomic Sdeace, 1966, 6, 277-278.

WEITZMAN, E. D., KRIPKE, D. F., GOLDMACHER, D., McGREGOR, P., & NOGEIRE, C. Acute reversal of thc sleep-waking cYcle in man: Effect on sleep stage patterns. Archives of Neurology, 1970, 22, 483-489. WILLIAMS, H. L., HAMMACK, J. T., DALY, R. L., DEMENT, W. C., & LU BIN, A. Responses to auditory stimulation, sleep loss and the EEG stages o f sleep. Electroencephalography & Clinical N europhysiology, 1964, 16, 269-279. WILLIAMS, R. L., AGNEW, H. W., JR., & WEBB, W. B. Sleep patterns in young adults: An EEG study. Electroencephalography & Clinical Neurophysiology, 1964,17,376-381. WILLIAMS, R. L., AGNEW, H. W., JR., & WEBB, W. B. Sleep patterns in the young adult female: An EEG study. Electroencephalography & Clinical Neurophysiology, 1966,20,264-266. WlNER, B. J. Statislieal prineiples in experimental design. New York: McGraw-Hill,1962.

The effects of practice upon rehearsal in short-term memory* EARL C. BUTTERFIELD and DEBORAH J. PELTZMAN University of Kansas Medical Center, Kansas City, Kans. 66103 and JOHN M. BELMONT, Yale University, New Haven, Conn. 06150 Ss were allowed to pace their acquisition of nine items in a memory task, and the distribution of their rehearsal was measured. Peak rehearsal pauses changed positions across the first five trials but occurred at the same positions across the last five. Thus, Ss used the first trials to test various acquisition strategies and select among them the one that enabled them to maximize their recall. Contemporary theories of memory and explanations of experimental findings are relying increasingly on the postulation of cognitive processes rather than associative concepts (Norman, 1970). Consider Restle's • The research reported here was supported by USPHS Grants HD-00183 and HD-04760. Reprint requests should be sent to the first author at 3933 Eaton Street Kansas City, Kans. 66103. '

Psychon. Sei., 1971, Vol. 23 (4)

(1970) explanation of his transfer-of-training findings. He pretrained two groups of Ss in a continuous paired-associates task, one group having relatively long presentation-to-test intervals (Group 1) and the other experiencing shorter intervals (Group 2). Both groups were then tested on the same task with variable intervals. Restle found that the group trained with

longer memory intervals showed greater accuracy on the test trials over all intervals. He suggested that Ss' information-processing strategies were a function of the demands of the task on wh ich they were trained and that once a cognitive structure was learned in training, it would be transferred to the test trials. Group 2 Ss could perform adequately in their training trials by merely storing a few recency items in a short-term rehearsal buffer. However, Group 1 Ss had to employ a more permanent long-term code in training, which, when transferred to the test trials, presumably accounted for their superior performance at all intervals. Restle extrapolated from these findings the hypo thesis that the phenomenon of improvement over trials in a memory task, commonly known as learning-to-learn or practice effects, might reflect the fact that it takes several trials for Ss to select the most efficient strategy for the task at hand rather than the refinement across trials of any one strategy. The present experiment was designed to test Restle's hypothesis that Ss use the first few trials of a short-term memory task to select an appropriate acquisition strategy. The experiment uses a multiple-list, single-trial, single-probe serial memory task in which the Ss themselves determine the rate at wh ich the lists' items are presented. Previous investigations using this task have provided Ss with from 5 to 10 "practice" trials to familiarize themselves with the task (Belmont & Butterfield, 1969). Analyses of the Ss' interitem pauses have revealed that during the postpractice trials, Ss typically pause to rehearse near the middle of the lists but do not pause appreciably toward the end of the lists. The explanation for the use of this strategy is that recall of the probe from the early and middle portions of the lists requires that these items be transferred to a relatively permanent secondary memory store by means of rehearsal, which is accomplished du ring pauses near or just past the middle of the lists, whereas tbe terminal list items can be recalled accurately simply by committing tbem to an echo box or primary memory store. If Ss adopt this sort of strategy gradually, then an examination of data from the first few trials, which are usually administered just to familiarize Ss with the task, should reveal evolving patterns of hesitations which gradually change to a stable pattern of pausing most just past the middle of the lists. SUBJECfS The Ss were 11 undergraduate students who volunteered their services for $2.00. 275